A large-scale network multi-cloud virtual machine generation and scheduling method and system for a network target range

Abstract

The application discloses a large-scale network multi-cloud virtual machine generation scheduling method and system of a network shooting range, relates to the network shooting range and resource scheduling technology, and comprises the following steps: according to the resource condition of a current network scene and the historical record of virtual machine generation, synchronously or recycling the image cache of a local computing node on demand; collecting the image disk information stored in the local computing node, and recording the resource condition of the virtual machine generated by each image; converting the original resource data of each computing node into standardized available resource information, and completing the construction of a resource pool; according to a preset resource reservation strategy and the resource application condition of the current virtual machine generation, performing comprehensive weight calculation and sorting on each candidate computing node; and according to the weight calculation and sorting result, selecting the computing node, and issuing a virtual machine scheduling instruction to the cloud to which the computing node belongs. The application is used for solving the problems of incomplete resource sensing and unreasonable allocation of the existing scheduling scheme.

Description

Technical Field

[0001] This application relates to the fields of network range and resource scheduling technology, and in particular to a method and system for generating and scheduling large-scale network multi-cloud virtual machines in a network range. Background Technology

[0002] In the process of building large-scale network scenarios in a network test range, the scheduling efficiency, resource utilization, and cross-cloud adaptability of virtual machines directly determine the simulation effect and operational stability of the test range. Current mainstream virtual machine scheduling solutions in the industry face several practical challenges when dealing with large-scale deployment requirements, specifically as follows: From the perspective of resource allocation, existing scheduling schemes lack a comprehensive understanding of the resource status of multiple cloud computing nodes. They often adopt fixed allocation or simple round-robin strategies, resulting in some computing nodes operating under high load for extended periods, leading to problems such as excessive CPU and memory usage and response latency. Meanwhile, other nodes remain idle, resulting in a serious waste of hardware resources and low overall resource utilization, which cannot meet the needs of large-scale concurrent deployment of virtual machines.

[0003] In terms of cross-cloud scheduling capabilities, existing solutions are mostly designed for single cloud environments and lack a unified management mechanism for multi-cloud resources. Computing nodes on different cloud platforms cannot achieve collaborative scheduling. When the test range needs to deploy virtual machines across clouds to meet the needs of complex scenario simulation, it cannot achieve unified allocation and scheduling of resources, resulting in extremely poor flexibility and limiting the ability of the network test range to expand to large-scale scenarios.

[0004] From the perspective of virtual machine generation efficiency, existing solutions generally lack image caching management mechanisms. When generating and running virtual machines, disk image information needs to be read and written from shared storage via the network. Due to network bandwidth limitations, the image transmission latency is high, which not only reduces the speed of virtual machine creation, but also easily causes virtual machine startup failure due to network bottlenecks. It cannot meet the needs of rapidly generating tens of thousands of virtual machines in large-scale network scenarios.

[0005] From an operational perspective, the existing solution lacks a robust resource monitoring mechanism and cannot capture abnormal situations in the computing node status in real time. When problems such as resource unavailability or network interruption occur, it cannot be detected and handled in a timely manner, which can easily lead to virtual machine scheduling failures and test range operation interruptions, affecting the normal construction of the scenario.

[0006] Furthermore, the existing scheduling decision-making process lacks a scientific multi-dimensional evaluation system, and only selects nodes based on a single resource indicator. It cannot combine multiple factors such as virtual machine image profiles and resource reservation requirements to make comprehensive decisions, resulting in inaccurate scheduling decisions and further reducing scheduling efficiency and resource utilization. Summary of the Invention

[0007] This application provides a method and system for generating and scheduling large-scale multi-cloud virtual machines in a network test range. It solves the problems of incomplete resource awareness and unreasonable allocation in existing scheduling schemes, realizes real-time collection and integration of resource status of multiple cloud computing nodes, and balances the load of each computing node through the decision mechanism of this application, improves the overall resource utilization rate, and avoids the coexistence of resource idleness and overload.

[0008] This application provides a method for generating and scheduling large-scale multi-cloud virtual machines in a network test range, including: Based on the current network resource situation and the virtual machine generation history, synchronize or reclaim the local image cache of the compute node as needed; The system acquires node information of multiple cloud computing nodes, collects image disk information in the local storage of the computing nodes, and records the resource status of virtual machines generated by each image to create a profile of the image resources. Based on node information, mirror disk information, and mirror resource profiles, and according to preset resource configuration allocation coefficients, the raw resource data of each computing node is converted into standardized available resource information to complete the construction of the resource pool. Based on the preset resource reservation strategy and the resource request conditions generated by the current virtual machine, the resource pool is filtered in multiple dimensions, and the comprehensive weight calculation and sorting of each candidate computing node are performed. Based on the weight calculation and sorting results, select the compute node and send virtual machine scheduling instructions to the cloud to which the compute node belongs.

[0009] This application provides a large-scale network multi-cloud virtual machine generation and scheduling system for a network range, including a processor and a memory. The memory stores a computer program, and when the computer program is executed by the processor, it implements the steps of the aforementioned large-scale network multi-cloud virtual machine generation and scheduling method for a network range.

[0010] The method in this application can solve the problems of incomplete resource perception and unreasonable allocation in existing scheduling schemes, realize the real-time collection and integration of resource status of multiple cloud computing nodes, and balance the load of each computing node through the decision-making mechanism of this application, improve the overall resource utilization rate, and avoid the coexistence of resource idleness and overload.

[0011] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0012] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of this application. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 This is a schematic diagram of the architecture of the large-scale network multi-cloud virtual machine generation and scheduling method for network ranges in this application embodiment; Figure 2 This diagram illustrates the relationship between the resource awareness layer modules of the large-scale network multi-cloud virtual machine generation and scheduling method in the network range of this application embodiment. Figure 3 This is a schematic diagram of the architecture of a large-scale network multi-cloud virtual machine generation and scheduling system for a network range, as described in this application embodiment. Detailed Implementation

[0013] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0014] This application provides a method for generating and scheduling large-scale multi-cloud virtual machines in a network test range. The architecture of this method is as follows: Figure 1 As shown, the scheduling architecture corresponding to this scheduling method adopts a layered architecture design, which is divided into a monitoring and assurance layer, a resource awareness layer, a decision analysis layer, and a scheduling execution layer. Each layer works together to achieve efficient and stable scheduling and generation of multi-cloud virtual machines in large-scale scenarios. The scheduling method of this application includes the following steps: In step S101, based on the current network resource situation and the historical data generated by the virtual machine, the local image cache of the compute node is synchronized or reclaimed as needed. Specifically, corresponding to the monitoring and assurance layer, which is mainly responsible for ensuring resource availability, it includes an image cache management module and a resource availability monitoring module. The image cache management module will automatically synchronize or reclaim the local image cache of the compute node as needed, based on the current resource situation and the historical data generated by the virtual machine.

[0015] The resource availability monitoring module monitors the resource status in real time. If an unavailable resource is found, it will be reported to the resource information management module to ensure that the resource is always available and to guarantee the reliability of resource scheduling.

[0016] In step S102, node information of multiple cloud computing nodes is acquired, image disk information in the local storage of the computing nodes is collected, and resource information of virtual machines generated by each image is recorded to create a profile of the image resources. The resource awareness layer is the data support layer of the entire method, which includes a resource information acquisition module, a local storage information acquisition module, a virtual machine resource analysis module, and a resource information management module.

[0017] The resource information acquisition module is responsible for acquiring information from multiple cloud computing nodes, providing basic resource data for subsequent scheduling decisions. The local storage information acquisition module collects image disk information from the local storage of computing nodes, clarifying the image resource status of local storage. The virtual machine resource analysis module records the resource status of virtual machines generated by each image, creating a profile of the image resources to help the system match resources more accurately. The resource information management module is responsible for managing various types of resource information, integrating and storing the collected resource data to provide comprehensive resource data support for the decision analysis layer. The specific execution flow of this stage is as follows: Figure 2 As shown.

[0018] In step S103, based on node information, mirror disk information, and mirror resource profiles, and according to preset resource configuration allocation coefficients, the original resource data of each computing node is converted into standardized available resource information, thus completing the construction of the resource pool. Correspondingly, the decision analysis layer, as the core decision-making link of the entire scheduling process, includes a resource pool construction module, a resource pool filtering module, a resource pool weight calculation module, a resource pool sorting module, and a resource election module.

[0019] Resource Pool Construction Module: This module receives resource data synchronized from the resource information management module in the resource awareness layer. This data includes the number of CPU cores, memory capacity, storage capacity, network bandwidth, local image cache information, and image resource profiles for each available computing node. Based on preset resource configuration allocation coefficients, the module converts the raw resource data of each computing node into standardized available resource information, completing the construction of the resource pool and providing a foundation for subsequent selection and decision-making.

[0020] In step S104, the resource pool is filtered in multiple dimensions according to the preset resource reservation strategy and the resource request conditions generated by the current virtual machine, and the comprehensive weight calculation and sorting of each candidate computing node are performed.

[0021] In step S105, a compute node is selected based on the weight calculation and sorting results, and a virtual machine scheduling instruction is sent to the cloud to which the compute node belongs. In some embodiments, after sending the virtual machine scheduling instruction to the cloud to which the compute node belongs, the process further includes: generating the virtual machine on the corresponding cloud environment compute node, and providing feedback on the compute node creation status.

[0022] Specifically, corresponding to the scheduling and execution layer, which is responsible for executing instructions, it includes an instruction issuance module and a multi-cloud resource management module. The instruction issuance module issues virtual machine scheduling instructions to the cloud to which the selected compute node belongs. After receiving the instructions, the multi-cloud resource management module, based on the cloud environment identifier in the instructions, calls the virtual machine creation API of the corresponding cloud platform to complete the virtual machine creation on the specified compute node and feeds back the virtual machine creation status to the instruction issuance module. If instruction issuance fails, a retry mechanism is triggered to ensure that the virtual machine can be created successfully.

[0023] The method in this application can solve the problems of incomplete resource perception and unreasonable allocation in existing scheduling schemes, realize the real-time collection and integration of resource status of multiple cloud computing nodes, and balance the load of each computing node through the decision-making mechanism of this application, improve the overall resource utilization rate, and avoid the coexistence of resource idleness and overload.

[0024] In some embodiments, synchronizing or reclaiming the local image cache of the compute node on demand, based on the current network resource situation and the virtual machine generation history, includes: The usage frequency of each image is statistically analyzed, and based on the usage frequency and resource consumption, the images are dynamically synchronized to the local storage of the corresponding compute nodes; and, For images whose usage frequency is lower than a specified frequency threshold and whose storage occupancy exceeds a preset ratio threshold, they are reclaimed from the local storage of the compute node. This reduces the network transmission pressure during virtual machine creation, improves virtual machine creation speed, and reasonably controls the storage space status of the compute node.

[0025] In a specific example, the resource availability monitoring module monitors the resource status in real time. If an unavailable resource is detected, it is reported to the resource information management module to ensure that the resource is always available and to guarantee the reliability of resource scheduling.

[0026] A further resource pool construction module receives resource data synchronized from the resource information management module in the resource awareness layer. This data includes the number of CPU cores, memory capacity, storage capacity, network bandwidth, local image cache information, and image resource profile data for each available computing node. Based on preset resource configuration allocation coefficients, the module converts the raw resource data of each computing node into standardized available resource information, completing the construction of the resource pool. In some embodiments, multi-dimensional filtering of the resource pool is performed based on preset resource reservation policies and the resource request conditions generated by the current virtual machine, including: Perform hard-condition filtering to eliminate computing nodes whose available resources in any of the dimensions of CPU, memory, storage, and network bandwidth do not meet the minimum requirements for virtual machine generation; Perform resource reservation filtering. Based on the resource reservation rules of the test range, reserve a certain proportion of resources for emergency scenarios and eliminate computing nodes that do not meet the reservation strategy in order to select computing nodes that meet the conditions to build a candidate resource pool.

[0027] In some embodiments, calculating and ranking the comprehensive weights of each candidate computing node includes: The min-max normalization method is used to normalize the resource data for each dimension; Calculate the overall weight: Among them, among them, The overall weight of resources is [0, 100]; α is the weight coefficient of CPU dimension, β is the weight coefficient of memory dimension, γ is the weight coefficient of storage dimension, δ is the weight coefficient of network bandwidth dimension, ε is the weight coefficient of image cache dimension, and ζ is the weight coefficient of image portrait dimension. The values ​​of these coefficients are all in the range of [0,1]. Normalize the score for the CPU dimension. Normalize the score for the memory dimension. To normalize the score for the storage dimension, The network bandwidth normalization score, For the image cache normalization score, The mirror image is normalized to a score, and these scores all range from [0, 100]. After the calculation is completed, the resource pool weight calculation module outputs the comprehensive weight data of each candidate computing node to the resource pool sorting module.

[0028] The resources in the selected candidate resource pool are sorted from highest to lowest based on their overall weight. This is achieved using a resource sorting module, which receives the overall weight data of each candidate computing node, sorts the computing nodes in the candidate resource pool in descending order of their overall weight, and then outputs the sorted candidate resource pool list to the resource election module.

[0029] In some embodiments, the sorting process further includes: Based on the sorted list of candidate resource pools, an optimal computing node is selected by selecting nodes sequentially from the highest weight, taking into account the resource requirements generated by the current virtual machine. The resource election module sends the node information of the selected optimal computing node to the scheduling execution layer and updates the resource occupancy status in the resource pool, where the scheduling execution layer is used for instruction execution. In other words, after the election is complete, the resource election module sends the selected computing node information to the instruction issuing module of the scheduling execution layer, and simultaneously updates the resource occupancy status in the resource pool to ensure the accuracy of resource allocation.

[0030] In some embodiments, it also includes: During the monitoring process, the usage frequency of each image is counted. Based on the usage frequency and resource consumption, the images are dynamically synchronized to the local storage of the corresponding compute nodes. For images whose usage frequency is lower than a specified frequency threshold and whose storage consumption exceeds a preset ratio threshold, they are reclaimed from the local storage of the compute nodes.

[0031] This application also proposes a large-scale network multi-cloud virtual machine generation and scheduling system for a network test range, including a processor and a memory. The memory stores a computer program, which, when executed by the processor, implements the steps of the aforementioned large-scale network multi-cloud virtual machine generation and scheduling method for a network test range. Specifically, as follows... Figure 3 As shown, the large-scale network multi-cloud virtual machine generation and scheduling system of this application includes the following scheduling process: 1. Monitoring and Security Phase During this phase, the image cache management module calculates the usage frequency of each image based on the virtual machine's historical generation history. Combining usage frequency with resource consumption, it dynamically synchronizes images to the local storage of the corresponding compute nodes. Images with low usage frequency and excessive storage consumption are automatically reclaimed from the compute node's local storage. Simultaneously, the resource availability monitoring module monitors resource status in real time and reports unavailable resources to the resource information management module, ensuring the availability and health of computing resources.

[0032] 2. Resource Perception Stage Through the resource information collection module, local storage information collection module, and virtual machine resource analysis module, the system completes the collection of resource information from multiple cloud computing nodes and local storage resources, while also creating a profile of mirrored resources, providing comprehensive and accurate resource data support for the decision analysis phase.

[0033] 3. Decision Analysis Stage As the core decision-making link in the entire scheduling process, the resource pool is constructed, filtered, weighted, sorted and selected based on the full resource data provided by the resource awareness layer, and finally the scheduling decision instructions are output to the scheduling execution layer.

[0034] 1) The resource pool construction module receives resource data synchronized from the resource awareness layer, converts the raw resource data into standardized available resource information, and completes the construction of the resource pool.

[0035] 2) The resource pool filtering module performs multi-dimensional filtering on the initial resource pool. First, it removes computing nodes that do not meet the minimum resource requirements for virtual machine generation, and then it removes computing nodes that do not meet the resource reservation policy to obtain the candidate resource pool.

[0036] 3) The resource pool weight calculation module performs comprehensive weight calculation on the candidate computing nodes. First, it normalizes the resource data of each dimension, and then calculates the comprehensive weight of each node through the comprehensive weight formula.

[0037] 4) The resource pool sorting module sorts the computing nodes in the candidate resource pool according to the comprehensive weight.

[0038] 5) The resource election module selects the optimal computing node, sends the node information to the scheduling execution layer, and updates the resource occupancy status.

[0039] 4. Scheduling and Execution Phase The instruction issuing module issues virtual machine scheduling instructions to the cloud to which the selected compute node belongs. The multi-cloud resource management module receives the instructions and completes the generation of the virtual machine on the corresponding cloud environment compute node, while also providing feedback on the creation status.

[0040] The method in this application adopts a layered architecture design to solve the problems of chaotic architecture and unclear responsibilities in existing solutions. It clarifies the core responsibilities of the monitoring and protection layer, resource awareness layer, decision analysis layer, and scheduling execution layer, and enables each layer to work together. Compared with existing solutions, it significantly improves the maintainability and scalability of the architecture, facilitates subsequent function upgrades and module replacements, and solves the pain point that existing scheduling systems are difficult to adapt to subsequent iterative optimizations.

[0041] To address the issues of incomplete resource awareness, unreasonable allocation, and uneven load distribution in existing solutions, this solution comprehensively collects various resource information from multiple cloud computing nodes through a resource awareness layer. The decision analysis layer employs multi-dimensional weight calculation and optimal election strategies to achieve precise resource allocation and effectively balance the load of each computing node. Compared to existing fixed allocation and simple round-robin scheduling methods, this solution significantly improves overall resource utilization and completely resolves the coexistence of resource idleness and overload.

[0042] To address the issues of existing solutions lacking image cache management, virtual machine creation being limited by network bottlenecks, and inadequate resource monitoring and failure to handle anomalies in a timely manner, the image cache management module reduces network transmission pressure and significantly improves virtual machine creation speed by synchronizing and reclaiming image caches on demand, thus solving the pain point of low virtual machine creation efficiency in large-scale scenarios. At the same time, the resource availability monitoring mechanism captures and handles resource anomalies in real time, effectively ensuring the availability of scheduled resources and avoiding the problem of test range construction interruption compared to existing solutions.

[0043] This solution addresses the problem that existing solutions are mostly designed for single-cloud environments and cannot uniformly schedule cross-cloud resources. By using a multi-cloud resource management module, it enables collaborative scheduling of resources from different cloud platforms, breaking the limitations of a single cloud environment. Compared with existing solutions, it significantly improves the flexibility of resource scheduling, meets the complex scenario requirements of deploying virtual machines across clouds in network test ranges, and expands the scenario adaptability of test ranges.

[0044] Existing solutions rely on shared storage for virtual machine operation, are limited by network transmission, and cannot meet the needs of large-scale virtual machine generation. By using a local image caching mechanism, virtual machine operation is freed from the dependence on shared storage and is not limited by network bandwidth. The number of virtual machines generated can be horizontally scaled with the number of physical servers. Compared with existing solutions, it can easily meet the needs of rapid concurrent generation of tens of thousands of virtual machines in ultra-large-scale network scenarios, thus solving the scale bottleneck of existing solutions.

[0045] It should be noted that, in the embodiments of this application, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0046] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0047] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0048] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims. All of these forms are within the protection scope of this application.

Claims

1. A method for generating and scheduling large-scale multi-cloud virtual machines in a network test range, characterized in that, include: Based on the current network resource situation and the virtual machine generation history, synchronize or reclaim the local image cache of the compute node as needed; The system acquires node information of multiple cloud computing nodes, collects image disk information in the local storage of the computing nodes, and records the resource status of virtual machines generated by each image to create a profile of the image resources. Based on node information, mirror disk information, and mirror resource profiles, and according to preset resource configuration allocation coefficients, the raw resource data of each computing node is converted into standardized available resource information to complete the construction of the resource pool. Based on the preset resource reservation strategy and the resource request conditions generated by the current virtual machine, the resource pool is filtered in multiple dimensions, and the comprehensive weight calculation and sorting of each candidate computing node are performed. Based on the weight calculation and sorting results, select the compute node and send virtual machine scheduling instructions to the cloud to which the compute node belongs.

2. The method for generating and scheduling large-scale network multi-cloud virtual machines in a network test range as described in claim 1, characterized in that, Based on the current network resource situation and the virtual machine generation history, the local image cache of the compute node is synchronized or reclaimed on demand, including: The usage frequency of each image is statistically analyzed, and based on the usage frequency and resource consumption, the images are dynamically synchronized to the local storage of the corresponding compute nodes; and, For images whose usage frequency is lower than a specified frequency threshold and whose storage usage exceeds a preset ratio threshold, they are reclaimed from the local storage of the compute node.

3. The method for generating and scheduling large-scale network multi-cloud virtual machines in a network test range as described in claim 1, characterized in that, Based on the preset resource reservation policy and the resource request conditions generated by the current virtual machine, the resource pool is filtered in multiple dimensions, including: Perform hard-condition filtering to eliminate computing nodes whose available resources in any of the dimensions of CPU, memory, storage, and network bandwidth do not meet the minimum requirements for virtual machine generation; Perform resource reservation filtering. Based on the resource reservation rules of the test range, reserve a certain proportion of resources for emergency scenarios and eliminate computing nodes that do not meet the reservation strategy in order to select computing nodes that meet the conditions to build a candidate resource pool.

4. The method for generating and scheduling large-scale network multi-cloud virtual machines in a network test range as described in claim 3, characterized in that, The comprehensive weight calculation and ranking of each candidate computing node includes: The min-max normalization method is used to normalize the resource data for each dimension; Calculate the overall weight: Among them, among them, The overall weight of resources is [0, 100]; α is the weight coefficient of CPU dimension, β is the weight coefficient of memory dimension, γ is the weight coefficient of storage dimension, δ is the weight coefficient of network bandwidth dimension, ε is the weight coefficient of image cache dimension, and ζ is the weight coefficient of image portrait dimension. The values ​​of these coefficients are all in the range of [0,1]. Normalize the score for the CPU dimension. Normalize the score for the memory dimension. To normalize the score for the storage dimension, The network bandwidth normalization score, For the image cache normalization score, Normalize the score for the mirror image; Based on resource weights, the resources in the selected candidate resource pool are sorted from high to low according to their overall weight.

5. The method for generating and scheduling large-scale network multi-cloud virtual machines in a network test range as described in claim 4, characterized in that, The sorted result also includes: Based on the sorted list of candidate resource pools, an optimal computing node is selected by selecting nodes sequentially from the highest weight, taking into account the resource requirements generated by the current virtual machine. The node information of the selected optimal computing node is sent to the scheduling execution layer, and the resource occupancy status in the resource pool is updated, wherein the scheduling execution layer is used for instruction execution.

6. The method for generating and scheduling large-scale network multi-cloud virtual machines in a network test range as described in claim 1, characterized in that, Also includes: During the monitoring process, the usage frequency of each image is counted. Based on the usage frequency and resource consumption, the images are dynamically synchronized to the local storage of the corresponding compute nodes. For images whose usage frequency is lower than a specified frequency threshold and whose storage consumption exceeds a preset ratio threshold, they are reclaimed from the local storage of the compute nodes.

7. The method for generating and scheduling large-scale network multi-cloud virtual machines in a network test range as described in claim 1, characterized in that, After issuing the virtual machine scheduling instruction to the cloud to which the compute node belongs, the following is also included: The virtual machine is generated on the corresponding cloud environment computing node, and the computing node creation status is reported back.

8. A large-scale network multi-cloud virtual machine generation and scheduling system for a network test range, characterized in that, It includes a processor and a memory, wherein the memory stores a computer program, which, when executed by the processor, implements the steps of the method for generating and scheduling large-scale network multi-cloud virtual machines for a network range as described in any one of claims 1 to 7.

Images